[2310.03718] Constraint-Conditioned Policy Optimization for Versatile Safe Reinforcement Learning